In recent years, with the popularization of cloud computing, each company increasingly uses resources in data centers provided by cloud service business providers, instead of having company's resources, such as servers, storage, or the like. Furthermore, as the type of data centers, a case is also increased, in which data centers are installed across a plurality of countries instead of remaining in a single country.
Furthermore, in data centers, it is desirable to implement an outsourcing service (hereinafter, referred to as a “one stop service”) that has a function of providing various kinds of services at one stop without receiving geographical and physical constraints. The one stop mentioned here indicates a state in which various services can be received at a single location. To implement the one-stop service, for example, it is considered to use a system constituted on the basis of a concept in which service management, operation, service platform, facilities, networks, or the like are integrated and on the basis of standardized service management.
Furthermore, in the one-stop service, it is desirable to efficiently operate and manage systems distributed in a plurality of countries or regions. With this configuration, it is preferable to construct a mechanism in which the operation level of a data center is not decreased even at the time of deterioration in the security situation of the country in which a data center is operated.
As an outsourcing service provided by a data center, for example, there is a system in which a plurality data centers are accommodated and an outsourcing service provider determines, in accordance with a use mode of a customer, a data center to operate a service. The customer simply uses the service provided from the data center and does not need to have a consciousness that a virtual machine that provides the service is operated in which one of data centers.
Furthermore, in each of the data centers, in order to provide an efficient service, there may be a case of monitoring the state of each of the virtual machines and moving the service to an appropriate data center in accordance with the nature or the intended purpose of the customer or in accordance with the operation state of the data center. For one thing, a data center is moved due to customer's convenience. For example, if the main communication path for the customer is changed and the line path is reviewed, it is conceivable that, for convenience of communication path, the service is moved to the data center in which the line speed is advantageous. Furthermore, as another circumstance, the data center is moved due to provider's convenience. For example, if a new data center is established or if a data center that provides a service is reviewed due to a trouble, it is conceivable that the service is moved to the newly established data center or the other data center by taking into consideration the service level provided to the customer is not decreased. Hereinafter, a move of a service from a certain data center to the other data center is referred to as “migration”. Then, in both cases, it is important to flexibly and rapidly relocate data and switch the service to the data center at the migration destination.
Here, there may be a case in which, in the one-stop service, a plurality of data centers is present. In such a case, it is preferable that, in order to meet the demand of a customer, a service provider allocates the provided service to an appropriate data center and flexibly changes the data center in accordance with the circumstances.
Therefore, it is conceivable to use a method that secures data related to services provided by a certain data center in the other data center as a backup and, when migration is actually performed, by using the subject backup data, the service is moved to the other data center. Then, as a technique that backs up data between the data centers, there is a method of performing a normal backup in which a copy of data is acquired at certain intervals and there is another method of copying the same data content into the other disk. In a description below, the latter method is referred to as mirroring. There are two types of mirroring, i.e., the perfect synchronization mirroring in which, in order to obtain the same content of the original data, an update of data is not enabled until both the original data and the data in the mirroring destination have been completely updated and asynchronous mirroring in which an update of data is enabled without waiting for the completion of the update in the mirroring destination.
Here, as a backup technology of data, there is a conventional technology, in which metadata is written at the reference time point of a backup and, if a write occurs before the next reference time point, data is written in time series after the metadata. Furthermore, there is a conventional technology that sets an identifier that indicates the type of data in both the metadata targeted for the backup and the file and that determines a backup destination server in accordance with the identifier, and sends the file.
Patent Document 1: Japanese Laid-open Patent Publication No. 2010-49488
Patent Document 2: Japanese Laid-open Patent Publication No. 2005-92282
However, with a normal backup, a certain interval is present for the backup acquisition. Consequently, although the effect on the data update performance is small, a difference occurs in operation data depending on the time difference after the acquisition. Furthermore, in the synchronization mirroring, the completeness of data is guaranteed; however, a data update cost is high and the response performance of the system tends to be decreased. Furthermore, in the asynchronous mirroring, the effect on the data response performance is decreased; however, it is difficult to guarantee the completeness of data. In this way, the timing of the data synchronization between the backup source and the backup destination and the cost needed for a backup has a tradeoff relationship. Thus, in order to implement flexible migration, it is preferable to perform a backup at high speed in order to guarantee the completeness of the data while reducing the cost.
However, even if a high speed backup is implemented, if the backup destination is only one, the number of options of the migration destination is decreased. Thus, it is conceivable that a backup is performed in a plurality of places; however, the cost of the backup is increased in accordance with the increase in the backup destination.
Furthermore, even if the conventional technology that writes data in time series after the metadata that was written at the reference time point is used, because the backup source is operation data, if a backup is performed in a plurality of places, the load of the data center that provides the service is increased and it is difficult to reduce the cost. Furthermore, even if the conventional technology that determines the backup destination by using the identifier that indicates the type of data is used, because the backup source is operation data, if a backup is performed in a plurality of places, the load of the data center that provides the service is increased and it is difficult to reduce the cost.